Mutual Coupling in RIS-Aided Communication: Model Training and Experimental Validation

This paper investigates the mutual coupling in reconfigurable intelligent surface (RIS)-aided communication. Despite the existence of several mutual coupling-aware models for RIS-aided communication, the absence of experimental validation remains a notable gap. To fill this void, this work delves into the recently proposed mutual coupling-aware communication model based on scattering matrices. Utilizing a fabricated 1bit quasi-passive RIS prototype operating in the mmWave band, we first propose a novel model training approach based on a single 3D full-wave simulation of the RIS radiation pattern, enabling the estimation of the scattering matrix among RIS unit cells. Subsequently, the trained model is validated via both full-wave simulations and experimental measurements on the real RIS prototype. Compared with the conventional communication model without accounting for mutual coupling in RIS, the utilized mutual coupling-aware model, incorporating trained scattering parameters, showcases a more accurate prediction performance. Benchmarked against the full-wave simulated RIS radiation pattern, the trained model can reduce the prediction error by up to 10.6 %. Meanwhile, the S-parameter between the Tx and Rx antennas is measured, validating that the trained model exhibits a closer alignment with the experimental measurements. Such results affirm the correctness of the adopted model and the effectiveness of the proposed model training method.